The Paradox of Choice Read online

  What often saves us from our faulty decision-making process is that different people experience different vivid or salient events, and thus have different events available to memory. You may have just read that Kias are actually very safe and you are all set to buy one. You mention this to me, but I’ve just read a story about a Kia being crushed by an SUV in an accident. So I tell you about my vivid memory, and that convinces you to revise your opinion. We are all susceptible to making errors, but we’re not each susceptible to making the same errors, because our experiences are different. As long as we include social interactions in our information gathering, and as long as our sources of information are diverse, we can probably steer clear of the worst pitfalls.

  The benefits of multi-individual information assessment is nicely illustrated by a demonstration that financial analyst Paul Johnson has done over the years. He asks students to predict who will win the Academy Award in several different categories. He tabulates the predictions and comes up with group predictions—the nominees chosen by the most people for each category. What he finds, again and again, is that the group predictions are better than the predictions of any individual. In 1998, for example, the group picked eleven out of twelve winners, while the average individual in the group picked only five out of twelve, and even the best individual picked only nine.

  But while diversity of individual experience can limit our propensity to choose in error, how much can we count on diversity of experience? As the number of choices we face continues to escalate and the amount of information we need escalates with it, we may find ourselves increasingly relying on secondhand information rather than on personal experience. Moreover, as telecommunications becomes ever more global, each of us, no matter where we are, may end up relying on the same secondhand information. National news sources such as CNN or USA Today tell everyone in the country, and now even the world, the same story, which makes it less likely that an individual’s biased understanding of the evidence will be corrected by his friends and neighbors. Those friends and neighbors will have the same biased understanding, derived from the same source. When you hear the same story everywhere you look and listen, you assume it must be true. And the more people believe it’s true, the more likely they are to repeat it, and thus the more likely you are to hear it. This is how inaccurate information can create a bandwagon effect, leading quickly to a broad, but mistaken, consensus.

  Anchoring

  SENSITIVITY TO AVAILABILITY IS NOT OUR ONLY ACHILLES’ HEEL when it comes to making informed choices. How do you determine how much to spend on a suit? One way is to compare the price of one suit to another, which means using the other items as anchors, or standards. In a store that displays suits costing over $1,500, an $800 pinstripe may seem like a good buy. But in a store in which most of the suits cost less than $500, that same $800 suit might seem like an extravagance. So which is it, a good buy or a self-indulgence? Unless you’re on a strict budget, there are no absolutes. In this kind of evaluation, any particular item will always be at the mercy of the context in which it is found.

  One high-end catalog seller of mostly kitchen equipment and gourmet foods offered an automatic bread maker for $279. Sometime later, the catalog began to offer a larger capacity, deluxe version for $429. They didn’t sell too many of these expensive bread makers, but sales of the less expensive one almost doubled! With the expensive bread maker serving as an anchor, the $279 machine had become a bargain.

  Anchoring is why department stores seem to have some of their merchandise on sale most of the time, to give the impression that customers are getting a bargain. The original ticket price becomes an anchor against which the sale price is compared.

  A more finely tuned example of the importance of the context of comparison comes from a study of supermarket shoppers done in the 1970s, shortly after unit-pricing started appearing on the shelves just beneath the various items. When unit price information appeared on shelf tags, shoppers saved an average of 1 percent on their grocery bills. They did so mostly by purchasing the larger-sized packages of whatever brand they bought. However, when unit prices appeared on lists comparing different brands, shoppers saved an average of 3 percent on their bills. They did so now mostly by purchasing not larger sizes, but cheaper brands. To understand the difference, think about how most supermarket shelves are arranged. Different-sized packages of the same brand are typically adjacent to each other. In this case, what the shopper gets to see, side by side, is the “small,” “large,” and “family” sizes of the same item along with their respective unit prices. This makes it easy to compare unit prices within the same brand. To compare unit prices across brands might require walking from one end of the aisle to the other. The multibrand list of unit prices makes it easier for shoppers to do cross-brand comparisons. And when such comparisons are easy to make, shoppers follow through and act on the information.

  When we see outdoor gas grills on the market for $8,000, it seems quite reasonable to buy one for $1,200. When a wristwatch that is no more accurate than one you can buy for $50 sells for $20,000, it seems reasonable to buy one for $2,000. Even if companies sell almost none of their highest-priced models, they can reap enormous benefits from producing such models because they help induce people to buy their cheaper (but still extremely expensive) ones. Alas, there seems to be little we can do to avoid being influenced by the alternatives that anchor our comparison processes.

  Frames and Accounts

  AND CONTEXT THAT INFLUENCES CHOICE CAN ALSO BE CREATED BY language.

  Imagine two gas stations at opposite corners of a busy intersection. One offers a discount for cash transactions and has a big sign that says:

  DISCOUNT FOR PAYING CASH!

  CASH—$1.45 per GALLON

  CREDIT—$1.55 per GALLON

  The other, imposing a surcharge for credit, has a small sign, just above the pumps, that says:

  Cash—$1.45 per Gallon

  Credit—$1.55 per Gallon

  The sign is small, and doesn’t call attention to itself, because people don’t like surcharges.

  Beyond the difference in presentation, though, there is no difference in the price structure at these two gas stations. A discount for paying cash is, effectively, the same as a surcharge for using credit. Nonetheless, fuel-hungry consumers will have very different subjective responses to the two different propositions.

  Daniel Kahneman and Amos Tversky call this effect framing. What determines whether a given price represents a discount or a surcharge? Consumers certainly can’t tell from the price itself. In addition to the current price, potential buyers would need to know the standard or “reference” price. If the reference price of gas is $1.55, then those who pay cash are getting a discount. If the reference price is $1.45, then those who use credit are paying a surcharge. What the two gas station proprietors are offering is two different assumptions about the reference price of gas.

  The effects of framing become even more powerful when the stakes are higher:

  Imagine that you are a physician working in an Asian village, and six hundred people have come down with a life-threatening disease. Two possible treatments exist. If you choose treatment A, you will save exactly two hundred people. If you choose treatment B, there is a one-third chance that you will save all six hundred people, and a two-thirds chance that you will save no one. Which treatment do you choose, A or B?

  The vast majority of respondents faced with this choice choose treatment A. They prefer saving a definite number of lives for sure to the risk that they will save no one. But now consider this slightly different problem:

  You are a physician working in an Asian village, and six hundred people have come down with a life-threatening disease. Two possible treatments exist. If you choose treatment C, exactly four hundred people will die. If you choose treatment D, there is a one-third chance that no one will die, and a two-thirds chance that everyone will die. Which treatment do you choose, C or D?

  Now the overwhelming ma
jority of respondents choose treatment D. They would rather risk losing everyone than settle for the death of four hundred.

  It seems to be a fairly general principle that when making choices among alternatives that involve a certain amount of risk or uncertainty, we prefer a small, sure gain to a larger, uncertain one. Most of us, for example, will choose a sure $100 over a coin flip (a fifty-fifty chance) that determines whether we win $200 or nothing. When the possibilities involve losses, however, we will risk a large loss to avoid a smaller one. For example, we will choose a coin flip that determines whether we lose $200 or nothing over a sure loss of $100.

  But the fact of the matter is that the dilemma facing the physician in each of the two cases above is actually the same.

  If there are six hundred sick people, saving two hundred (choice A in the first problem) means losing four hundred (choice C in the second problem). A two-thirds chance of saving no one (choice B in the first problem) means a two-thirds chance of losing everyone (choice D in the second problem). And yet, based on one presentation, people chose risk, and based on the other, certainty. Just as in the matter of discounts and surcharges, it is the framing of the choice that affects our perception of it, and in turn affects what we choose.

  Now let’s look at another pair of questions:

  Imagine that you have decided to see a concert where admission is $20 a ticket. As you enter the concert hall, you discover that you have lost a $20 bill. Would you still pay $20 for a ticket to the concert?

  Almost 90 percent of respondents say yes. In contrast:

  Imagine that you have decided to see a concert and already purchased a $20 ticket. As you enter the concert hall, you discover that you have lost the ticket. The seat was not marked and the ticket cannot be recovered. Would you pay $20 for another ticket?

  In this situation, less than 50 percent of respondents say yes. What is the difference between these two cases? From the perspective of the “bottom line,” they appear the same; both involve a choice between seeing a concert and being $40 poorer or not seeing it and being $20 poorer. Yet obviously we don’t seem to see them as the same, because so many respondents choose differently in the two cases. Kahneman and Tversky suggest that the difference between the two cases has to do with the way in which we frame our “psychological accounts.” Suppose that in a person’s psychological ledger there is a “cost of the concert” account. In the first case, the cost of the concert is $20 charged to that account. But the lost $20 bill is charged to some other account, perhaps “miscellaneous.” But in the second case, the cost of the concert is $40; the cost of the lost ticket, plus the cost of the replacement ticket, both charged to the same account.

  The range of possible frames or accounting systems we might use is enormous. For example, an evening at a concert could be just one entry in a much larger account, say a “meeting a potential mate” account, because you’re going out in the hope of meeting someone who shares your interests. Or it could be part of a “getting culture” account, in which case it would be one entry among others that might include subscribing to public television, buying certain books and magazines, and the like. It could be part of a “ways to spend a Friday night” account, in which case it would join entries like hanging out at a bar, going to a basketball game, or staying home and dozing in front of the television. How much this night at a concert is worth will depend on which account it is a part of. Forty dollars may be a lot to spend for a way to fill Friday evening, but not much to spend to find a mate. In sum, just how well this $40 night at the concert satisfies you will depend on how you do your accounting. People often talk jokingly about how “creative” accountants can make a corporate balance sheet look as good or as bad as they want it to look. Well, the point here is that we are all creative accountants when it comes to keeping our own psychological balance sheet.

  Frames and Prospects

  KAHNEMAN AND TVERSKY HAVE USED THEIR RESEARCH ON FRAMING and its effects to construct a general explanation of how we go about evaluating options and making decisions. They call it prospect theory.

  If you look at the diagram above, you see objective states of affairs along the horizontal axis—positive to the right of the vertical axis, and negative to the left of it. These might be gains or losses of money, gains or losses of status on the job, gains or losses in your golf handicap, and so on. Along the vertical axis are subjective or psychological responses to these changes in states of the affairs. How good do people feel when they win $1,000 at the racetrack? How bad do people feel when their golf handicap goes up three strokes? If psychological responses to changes were perfectly faithful reflections of those changes, the curve relating the objective to the subjective would be a straight line that went right through the 0-point, or origin, of the graph. But as you can see, that is not the case.

  To figure out why prospect theory gives us this curve rather than a straight line, let’s look at the two halves of the graph separately. The top, right portion of the graph depicts responses to positive events. The thing to notice about this curve is that it’s steepness decreases as it moves further to the right. Thus, an objective gain of say $100 may give 10 units of subjective satisfaction, but a gain of $200 won’t give 20 units of satisfaction. It will give, say, 18 units. As the magnitude of the gain increases, the amount of additional satisfaction people get out of each additional unit decreases. The shape of this curve conforms to what economists have long talked about as the “law of diminishing marginal utility.” As the rich get richer, each additional unit of wealth satisfies them less.

  With the graph of prospect theory in view, think about this question: would you rather have $100 for sure or have me flip a coin and give you $200 if it comes up heads and nothing if it comes up tails? Most people asked this question go for the sure $100. Let’s see why. A sure $100 and a fifty-fifty chance for $200 are in some sense equivalent. The fact that the payoff for the risky choice is double the payoff for the sure thing exactly compensates for the fact that the chances you’ll get the payoff are halved. But if you look at the graph, you’ll see that psychologically, you won’t feel twice as good with $200 in your pocket as you will with $100 in your pocket. You’ll feel about 1.7 times as good. So to make the gamble psychologically worthwhile to you, I’d have to offer you something like $240 for a heads. Thus, Kahneman and Tversky point out, people tend to avoid taking risks—they are “risk averse”—when they are deciding among potential gains, potential positive outcomes.

  Now let’s look at the other side of the graph, which depicts response to losses. It too is a curve, not a straight line. So suppose I asked you this question: would you rather lose $100 for sure or have me flip a coin so that you lose $200 if it comes up heads and you lose nothing if it comes up tails? As in the last example, double the amount is compensated for by half the chances. If you don’t like risks in the first problem, you probably won’t like them in the second either. This suggests you’ll take the sure loss of $100. But chances are you didn’t, and the graph tells us why. Notice that the curve falls steeply at the beginning and then gradually levels off. This reflects what might be called the “decreasing marginal disutility of losses.” Losing the first $100 hurts worse than losing the second $100. So although losing $200 may be twice as bad objectively as losing $100, it is not twice as bad subjectively. What that means is that taking the risk to perhaps avoid losing anything is a pretty good deal. Thus, as Kahneman and Tversky again point out, people embrace risk—they are “risk seeking”—in the domain of potential losses.

  There is another feature of the graph worth noting: the loss portion of the graph is much steeper than the gain portion. Losing $100 produces a feeling of negativity that is more intense than the feelings of elation produced by a gain. Some studies have estimated that losses have more than twice the psychological impact as equivalent gains. The fact is, we all hate to lose, which Kahneman and Tversky refer to as loss aversion.

  The last and crucial element to the graph is the loc
ation of the neutral point. This is the dividing line between what counts as a gain and what counts as a loss, and here, too, subjectivity rules. When there is a difference in price between cash and credit at the gas station, is it a discount for cash or a surcharge for credit? If you think it’s a discount for cash, then you’re setting your neutral point at the credit-card price and paying cash is a gain. If you think it’s a surcharge, then you’re setting your neutral point at the cash price, and using your credit card is a loss. So fairly subtle manipulations of wording can affect what the neutral point is and whether we are thinking in terms of gains or losses. And these manipulations will in turn have profound effects on the decisions we make—effects that we really don’t want them to have, since in an important sense, discounts and surcharges are just two ways of saying the same thing.

  In the same way, we give disproportionate weight to whether yogurt is said to be 5 percent fat or 95 percent fat free. People seem to think that yogurt that is 95 percent fat free is a more healthful product than yogurt that has 5 percent fat, not realizing, apparently, that yogurt with 5 percent fat is 95 percent fat free.

  Or suppose you are one of a large group of participants in a study and for your time and trouble, you are given either a coffee mug or a nice pen. The two gifts are of roughly equal value and randomly distributed—half of the people in the room get one, while the other half get the other. You and your fellow participants are then given the opportunity to trade. Considering the random distribution, you would think that about half the people in the group would have gotten the object they preferred and that the other half would be happy to swap. But in fact, there are very few trades. This phenomenon is called the endowment effect. Once something is given to you, it’s yours. Once it becomes part of your endowment, even after a very few minutes, giving it up will entail a loss. And, as prospect theory tells us, because losses are more bad than gains are good, the mug or pen with which you have been “endowed” is worth more to you than it is to a potential trading partner. And “losing” (giving up) the pen will hurt worse than “gaining” (trading for) the mug will give pleasure. Thus, you won’t make the trade.